DBScholar

Back to papers

Towards Pattern-aware Data Augmentation for Temporal Knowledge Graph Completion

Summary: Tackles entity/timestamp imbalance and model preferences in temporal KGC by introducing Booster, the first pattern-aware data augmentation that generates temporally and semantically consistent synthetic facts. Booster validates candidates via hierarchical triadic-closure scoring and uses two-stage, frequency-filtered training to mine hard samples and avoid false negatives, improving TKGC models by ~4.5%. (summarized by gpt-5-mini on Feb 09 2026)

Paper ID
h42c3eec31eb6b7f9
Venue
VLDB
Year
2025
Pagerank
5.4772833e-05
Overall Rank
7,645 | 48.60%
DOI
10.14778/3748191.3748216

Incoming Non-self Citations Over Time

Authors

BibTeX Citation

@article{zhang_vldb25,
        title = {{Towards Pattern-aware Data Augmentation for Temporal Knowledge Graph Completion}},
        author = {Zhang, Jiasheng and Ouyang, Deqiang and Liang, Shuang and Shao, Jie},
        journal = {PVLDB},
        series = {{VLDB} '25},
        volume = {18},
        number = {10},
        pages = {3573--3586},
        doi = {10.14778/3748191.3748216},
        url = {https://doi.org/10.14778/3748191.3748216},
        year = {2025}
}

Incoming Citations (Sorted by Pagerank)

Showing 1 of 1 citing papers.

Previous Page 1 / 1 Next

Outgoing Citations (Sorted by Pagerank)

Showing 2 of 2 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Previous Page 1 / 1 Next

Semantically Similar Papers